Propagation Effects in Satellite Mounted RadarRemote Sensing

(Invited Paper)

Andreas DanklmayerDLR e.V. (German Aerospace Center)

Microwaves and Radar Institute82234 Wessling, Oberpfaffenhofen

GermanyEmail: Andreas.Danklmayer@dlr.de

Abstract— Space-borne Synthetic Aperture Radar (SAR) imag-ing is often considered to possess both day/night and all weatheroperational capabilities. However, as the operating frequenciesof SAR-systems are increasing, visible image distortions due toheavy precipitation in SAR-images may become present. Thisholds especially for the case of convective rain events imagedat X-band frequencies and beyond. These rain-cell signaturesare thoroughly investigated, and the physical background ofthe related propagation effects is provided. A review of raincell signatures from former missions like SIR-C/X-SAR andthe Shuttle Radar Topography Mission is given. Furthermore,the German spaceborne satellite TerraSAR-X delivered severalmeasurements, which facilitate to study precipitation effects inSAR images. Based on these SAR images and simultaneouslyacquired weather radar measurements, a quantitative estimationof precipitation effects in SAR images is presented. In a furtherstep, an attempt is made to extrapolate the effects observedin X-band SAR images to images acquired at higher nominalfrequency bands such as Ka-band.

I. INTRODUCTION

The philosophy and intention of this article is to give first anintroduction to the basic image formation concept of SyntheticAperture Radar (SAR) and its diverse applications. Thenwe will focus on the specific aspects related to microwavepropagation in the atmosphere. This will hopefully make thepresented material beneficial for young scientists entering tothe field. Experts from other domains not familiar with thebasic concepts will hopefully also gain from the introduction,which is necessary in order to understand the context for themore recent findings treated afterwards. In the following weitemize the aspects which are covered in this tutorial article:

• Synthetic Aperture Radar basic image formation conceptand applications

• Review of relevant propagation effects• Probability of rain induced effects in X-band• Physical interpretation of rain cell signatures in SAR

images• Modelling of rain cell signatures in SAR images• Short summary and conclusions

II. ESSENTIAL BASICS OF SYNTHETIC APERTURE RADAR

Synthetic Aperture Radar represents a two-dimensionalradar imaging technique based on active remote sensing of

the Earth or other planets. In contrast to optical sensors,SAR is independent of day- and nighttime and can penetratethrough clouds. However, as will be shown, SAR can beaffected by propagation effects through heavy rain underadverse weather conditions. Nowadays, SAR is widely usedfor remote sensing of the Earth using air- and spaceborneplatforms. Storage of the phase history allows for synthesisof a large synthetic aperture which improves the resolutionin the azimuth direction tremendously. This can be illustratedby a simple example. A space-borne system operating in X-band with an antenna length of 12 m, and altitude of 800 kmwould deliver only a poor resolution of 2000 m without usingthe synthetic aperture radar principle. However, with SAR thebest achievable resolution that can be obtained is D/2, whereD is the anntenna length. This result may seem astonishingsince the resolution is independent of range.

EarthquakesEarthquakes VolcanoesVolcanoes

OceanOcean

Land & Sea IceLand & Sea Ice

SubsidenceSubsidence

TrafficTraffic

Land EnvironmentLand Environment

ReconnaissanceReconnaissanceDisasterDisaster

Fig. 1: A selection of applications for SAR (courtesy of G.Krieger).

The received radar signals contain information about rough-ness and geometrical structure as well as electrical propertiesof the imaged scene. Furthermore the signal may deliverinformation of the atmosphere and surface. Dependent on thewavelength also information of the subsurface can be gained.

y

x

z

range

azimuth

antenna

illuminated

area

stationary

point target

beam

z

y

h

swath

width

far range

nearra

nge

antenna

qD

qD … depression angle

qi … incidence angle

r … slant range

r qi

swath

swath width

Fig. 2: A depiction of a SAR sidelooking imaging geometry.

Radar imaging has been proven to be very successful for awide range of remote sensing applications and some of themare exemplarily given in Fig. 1.

A typical SAR imaging geometry is given in Fig. 2, whichshows a side-looking antenna mounted on a moving platform.Due to their higher altitude, satellite incidence angles vary lessthan airborne incidence angles. This leads to more uniformillumination on satellite images than for airborne radar images.We emphasize, that the velocity v of the platform carryingthe antenna is small in comparison to the transmitted signals,which are propagating at a constant velocity equal to the speedof light (c ≈ 3 × 108 m/s). The antenna beam is pointingtowards the earth and is illuminating the surface within theantenna footprint (intersection of beam with surface). Theshaded area at the ground depicted in Fig. 2 is sometimescalled the antenna footprint.

Note that the beam width is inversely proportional to thedimension of the antenna and the frequency, i.e. the larger theantenna and the higher the frequency, the smaller the beamwidth. The width of the so-called swath (ranging from near- tofar range) limits the area of the surface which can be resolvedin a processed SAR image in the across track dimension. Theintersection of the vertical from the platform with the groundis called the nadir. In Fig. 2 the meaning of the depression-and incidence angle is shown.

The axis along the line of flight is called along-track orazimuth and the perpendicular direction is called in generalrange or across-track, where slant range and ground rangehave to be distinguished. The slant range denotes the distancefrom the aircraft to the scatterers, and the ground range thedistance from the nadir to the observed scatterer. The termsquint angle denotes the deviation angle between the mainaxis of the antenna beam and the plane perpendicular to theline of flight, where positive squint angles are by conventionassigned to positive Doppler frequency shifts, and vice versa.We note that along-track and range directions are orthogonalto each other.

A. Image Formation Process

As the platform moves along the azimuth direction thepulses are transmitted by the antenna with a certain pulserepetiton frequency (PRF). Instead of a short sharp pulse, a

Range FFT

Range IFFT

AzimuthFFT

AzimuthIFFT

x x

Signal Domain(Start)

Azimuth matchedfilter

Range matchedfilter

(End)Processed SAR imagewith complex pixel data.

S(x,y)

F(ù)

F( (ù)·G ù)R(t)(Rangecompressed)

S(r,ù)A(ù)

S( (r,ù)·A ù)

G(ù)

f(t)

Fig. 3: A flow chart of range Doppler imaging.

pulse having a large time-bandwidth product is transmitted.For obtaining a small range resolution and the highest possiblesignal-to-noise ratio pulse compression is carried out usinga matched filter. The output or focused signal, respectively,is given by the convolution of the received signal with thereference signal/function. Following the optimal filter theory,the reference or matched filter function, respectively, is thecomplex conjugated and time-reverted function of the SARsystem response.

The data collected by SAR systems are represented by atwo-dimensional complex matrix in order to be processed toobtain a two-dimensional representation of the illuminatedarea.

An appropriate algorithm in order to understand the process-ing procedure is called either Range-Doppler or rectangularimaging [1] described in the following. The flow chart depictedin Fig. 3 shows the main required steps to transform from theraw time signal domain into the complex image domain.

The basic procedure of compressing the raw data in orderto obtain an image is a two-dimensional correlation with areference function, which is a matched filter operation.

The starting point for image processing after acquisition andsampling of the received analog data is the two-dimensionalraw data matrix containing complex entries. The fundamentalalgorithm called ’rectangular’ algorithm is briefly discussed.The algorithm is based on the rectangular assumption, as-suming that the order of processing in range and azimuth isirrelevant and can be executed separately. The principle ofcompressing the raw data in range and azimuth via matchedfilter operations represents a well established and fundamentalprocessing algorithm. We limit the treatment of processingalgorithms herein. A variety of references are available cf.[1]–[6] that detail the subject more broadly.

Being a well established and matured remote sensing tech-nique, a large body of literature exists for various topics ofSAR imaging; reference is given to [7]–[11].

III. ASSESSMENT OF PROPAGATION EFFECTS IN SARIMAGES

Amongst the many effects that may affect SAR imagesare delays, attenuation, noise, scintillation which are causedby atmospheric gases, rain (precipitation), clouds, fog or freeelectrons in the ionosphere.

Whether these effects are relevant depends on the signalparameters, the path geometry and the conditions prevailingin the ionosphere and the troposphere. The use of higherfrequencies (X-band and beyond) minimises ionospheric ef-fects on propagation, but tropospheric effects often increaseor dominate as will be investigated in more detail.

The question to what extent atmospheric effects are criticaldepends also on the specific SAR application. For instance,for Synthetic aperture radar (SAR) interferometry, a powerfulremote sensing technique for measuring the distance to theEarth’s surface, the phase measurements is exploited. It re-quires that the surface is imaged by the radar at least twicefrom almost the same position in space. The differences inphase between the two images produce interferometric fringesthat allows to deduce the topography of the surface. Anyeffect on the phase of the SAR signal will affect the accuracyof the height information. The focus in this contributionis mainly to the propagation effects that cause attenuationand backscattering in the image. For atmospheric effects oninterferometric SAR techniques we refer to [12]–[14].

In the following some examples are shown how precipitationmay affect SAR intensity images. The first example (Fig. 4) istaken from the SIR-C/X-SAR mission flown in 1994, wherethree images acquired in three different frequency bands arecombined. The RGB-composite coding is (R) for X-band (G)for C-band, and (B) for L-band. As expected the ”blocking”is most pronounced for X-band frequencies.

Three further examples of rain affected signatures in SARimages are given in Fig. 5 all of them taken from areasover tropical rain-forest, where heavy precipitation possesseshighest probability on a world-wide scale.

For a recent measurement taken with TerraSAR-X, theGerman SAR satellite, launched in June 2007 and operatingextremly successful since then [15], Fig. 6 is provided.

The physical processes are better elucidated with the helpof Fig. 7 where a SAR amplitude cutout of cross-track profilethrough a rain cell is depicted. As diagrammatically shown, thebackscattering is accompanied with higher amplitude valuesand the weak signals behind the first maxima belongs to theblack shadows easily recognised in Fig. 7, where up to 20 dBdifference in the dB level may be observed.

IV. TROPOSPHERIC EFFECTS

The troposphere, as the lowest part of the Earth’s at-mosphere, reaches from the surface to approximately 12 kmabove ground and causes, amongst other effects, attenuationof traversing signals due to hydrometeors (rain, snow, hail),atmospheric gases, fog and clouds [16]. Except at low ele-vation angles, the attenuation of frequencies below 1 GHz isnegligible. Insignificant contributions to the attenuation willbe obtained for frequencies up to 10 GHz due to fog andnon-precipitating clouds. However the transmission spectrumexhibits peaks for frequencies around 22 GHz and 60 GHz dueto molecular resonances from gases, i.e. water vapour andoxygen. Whereas absorption effects due to atmospheric gasesare present constantly and everywhere, attenuation due to

X-band C-band L-band

35,3 km

21,3

km

Colour composite RGB - image

X-band C-band L-band

35,3 km

21,3

km

Colour composite RGB - image

Fig. 4: Example of influence of rain on SAR intensity images,using the RGB composite corresponding to X-band (R), C-band (G), and L-band (B) measurements, respectively. Therain-induced ‘blocking‘, as expected, is most pronounced inthe X-band case. The images have been recorded during theX-SAR/SIR-C Mission conducted by DLR/NASA in 1994.The image is centered at 11.2◦ south latitude and 61.7◦ westlongitude, respectively. North is toward the upper left.

condensed water in the form of precipitation, clouds and fog isinfrequent and is limited to certain areas. Attenuation consistsof two physical processes: the reduction of the wave’s energydue to the heating of the water particles and, the scattering ofenergy away from the main direction of propagation.

A. Global Characteristics of Rain

On average, 50 % of the Earth’s surface is coverered byclouds. However the rain to cloud ratio is seldom larger than10 %. This means that the upper bound of 5 % on Earth’ssurface is receiving rain at a given time. Rain events are dom-inated by light rain events, however higher rain rates accountfor most of the total liquid water reaching the surface. Evenfor the areas with the highest rain rate, the percentage timein a year that exceeds 50 mm/hr is 0.1 % (annual probability)= 526 min. And 150 mm/hr is exceeded less than 0.01 % =53 min.

B. Modelling of Attenuation and Backscattering in SAR Im-ages

For the modelling of the attenuation and backscatteringeffects in SAR images, Fig. 8 is clarifying the underlyinggeometry [17]. The diagram provided at the bottom of Fig. 8shows the qualitative variation of the normalised radar crosssection (NRCS) due to the idealised rain cell. The detailedmodelling and calculation is provided in the following twosections.

One of the major problems affecting microwave and mil-limetre wave bands for terrestrial and space-borne radars is theattenuation through rain [16]. A convenient way to describethe rain intensity is the so called rainfall-rate or rain-rate givenin millimetres per hour. This quantity refers to a certain fluxof rain towards the surface of the Earth and may be measured

Fig. 5: Three examples of weather-corrupted SAR imagesfrom the SRTM mission (X-band: 9 GHz) from the Brazilianrain forest. The horizontal corresponds to range, and thevertical to the azimuth direction. The direction of illuminationfor all three images is given from the left- to the right-handside. (a) DT: 087.070 Scene: 740. (b) DT: 150.050 Scene: 820.(c) DT: 039.070 Scene: 740.

Range & Illumination

Azimuth

10 km

Fig. 6: An example of a SAR image with rain-cell signaturesrecorded with TerraSAR–X in the US close to Baton Rouge.The white shading to the left of the dark spot is due to directreflections from the rain region. The darkly shaded areas aredue to rain attenuated (blocked) signals from the ground.

e.g. by gauges or weather radars. A widely acceped empiricalrelation of the form

γ(x, t) = a ·Rb (1)

between specific attenuation γ(x, t) and rain rate R is used tocalculate the specific attenuation for a given rain rate [16],[18]. The parameters a and b are dependent on the radiofrequency, the raindrop size distribution, the polarization andother factors [18].

The total attenuation for a given instant of time can beobtained by adding up the specific attenuation along the pathof propagation using the following expression [16]

Fig. 7: Depiction of a slant-range reflectivity profile (“A-scope”) for the rain-cell cut from a very recent TerraSAR-Xmeasurement over a tropical rain forest in Brazil. Such datasets may, at some stage, enable estimation of rain rate oversuch isolated areas. This figure was inspired by the study ofRunge et al. [19] who analyzed reflectivity profiles of SRTMdata takes over a tropical rain forest.

illuminating beam

H · cot H ·tan

H

idealized rain cell volume90°

Radar pulse volume

look angle

Range

Range

NRCS

qualitative variation of

the NRCS due rain cell

volume effects

Rain cell length

Earth

surface

Fig. 8: A depiction of a SAR imaging scenario of an idealisedrain cell. In the diagram the qualitative variation of the NRCSdue to the rain cell volume effects is given. A combination ofbackscattering and attenaution can occur, where qualitativelythe backscattering due to rain is the minor effect and attenu-ation the major [17].

A(t) = 2

∫ h

0

γ(x, t)dx [dB] (2)

where

A(t) . . . total attenuation for given time instant tt . . . timeh . . . path lengthγ(x, t) . . . specific attenuationx . . . position along the path of propagation

The specific attenuation along the slant path of propagation

Fig. 9: A plot of the specific attenuation given in units ofdB/km versus the rain rate in units of mm/h for four differentfrequencies (1, 6, 10 and 35 GHz) after [21].

has to be known. However, detailed knowledge of the mediumthrough which the signal propagates is rather limited andthe temporal and spatial variation of the medium requireassumptions and some modelling. In the case of precipitationwe may have some idea about the thermodynamic phase(ice, water, melting band) but no precise information. Usingthe calculated values for the specific attenuation in X-bandprovided in diagram Fig. 9 for a 5 km long path through aheavy tropical convective rain (70 mm/h and more) suggestsa 20 – 30 dB two-way attenuation, which was confirmed bycomparing the backscattering coefficient of affected and non-affected region from data takes acquired by TerraSAR-X overBrazilien rain forest [20].

C. Modelling of the Attenuation under Rain Conditions forKa-band

For the calculations of the rain rate to cause visible attenua-tion in Ka-band SAR images, a simple model shown in Fig. 10[17], is used. A reasonable height of 4 km is assumed, as wellas a homogenous rain rate for the modelled cell. The incidenceangle of the propagating signals was chosen 30o. With the helpof (1) and by using the regression coefficients in Table I, thespecific attenuation was calculated. These values are given inTable II. Finally, the values for the two-way path attenuationfor different rain rates (5, 50 and 100 mm/h) can be found inTable III.

As a total two-way attenuation of 25 dB becomes visiblein SAR images in X-band it is of interest which rain rate isnecessary to cause such an attenuation for Ka-band (35 GHz)frequencies. To this end, the following equation is applied

γ(t) =A(t)

2 · Hcos(θ)

[dB/km], (3)

where the rain rate is found using (1) with the accordingparameters a and b for Ka-band

R =b

√γ(t)

a. (4)

Range

H/sin

H·cosH/cos

yminymax

H

lookangle

Fig. 10: The structure of an idealised rain cell used for thecalculation in Section IV-C.

TABLE I: Regression coefficients used for the calculation ofthe specific attenuation cf. (1)

DSDFrequency Mar. Palmer Joss Thunderst.

a b a b

X-Band (10 GHz) 0.0136 1.15 0.0169 1.076

Ka-Band (35 GHz) 0.268 1.007 0.372 0.783

First calculations using the parameters of the idealised raincell of 4 km and using 25 dB of total two-way attenuation,assuming the Marshall Palmer Parameters for Ka-band attenu-ation deliver a rain rate close to 10 mm/h. This simple exampledemonstrates that such rain rates are fully capable to distortKa-band SAR measurements to a visible extent. In Fig. 11 therange of values for attenuation and rain rate are extended andthe diagram shows the two-way attenuation versus the rain ratefor different incidence angles at Ka-band. Similar informationis provided for X-band frequencies in Fig. 12 for comparison.The aforementioned simple assumptions of a homogenous rain

TABLE II: Specific Attenuation [dB/km]

Rainrate Specific Attenuation[mm/h] 10 GHz 35 GHz

5 0.08 1.3150 1.22 7.95

100 2.4 13.69

TABLE III: Maximum attenuation for modelled rain cell[dB].

Rainrate Specific Attenuation[mm/h] 10 GHz 35 GHz

5 0.739 12.150 11.26 73.43

100 22.17 126.46

0

2

4

6

8

10

12

14

16

18

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39

Total Two-way attenuation [dB]

Rain

rate

[m

m/h

]

30° 40° 45° 50° 60°

Fig. 11: A diagram of the two-way path attenuation and thecorresponding rain rate for the modelled rain cell of 4 kmheight (Fig. 10) at Ka-band (35 GHz), for different look angles.

0

20

40

60

80

100

120

140

160

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39

Total Two-way attenuation [dB]

Ra

inra

te [

mm

/h]

30° 40° 45° 50° 60°

Fig. 12: A diagram of the two-way path attenuation and thecorresponding rain rate for the modelled rain cell (Fig. 10) atX-band (10 GHz), for different look angles.

cell with constant precipitation may be somehow optimistic,since a melting layer precipitation may severly increase thetotal path attenuation.

D. Consideration of Backscattering Effects in SAR Imaging

The radar response (backscattering) from hydrometeors isdetermined by the raindrop size, shape, density, orientation,and temperature. Furthermore, the backscattering depends onthe polarization of the wave interacting with the precipitationmedia. The theoretical concept that describes the scatteringfrom a dielectric sphere was established by Mie in 1908. Therelation to estimate the backscatter cross section of a volumeof small particles by assuming the well-known Rayleighapproximation (the diameters D of the raindrops are muchsmaller than the wavelength λ) is given as [23]

σb =π5

λ4|K|2

N∑

i=1

D6i (5)

where

K =m2 − 1

m2 + 2(6)

and m is the complex index of refraction of the scattering

particle. The summation term for a distribution of particlescan be given as

Z =

∫N(De)D

6edDe [mm6/m3], (7)

where Z is termed the reflectivity factor, De is the diameterof each droplet and N(De) is the number of droplets per unitvolume.

N(De) = N0e−λdDe , λd = 4.1 ·R−0.21 (8)

where N0 and λd are the parameters defining the dropsize distribution (DSD). For the special case of the Marshall-Palmer distribution the parameter N0 is given as 8000 in m−3

mm−1 and λd [mm−1] is related to the rainfall intensity R[mm/h] as shown in Eq. (8).

Note that Z is commonly given in logarithmic units accord-ing to

Z = 10 · log10Z [dBZ]. (9)

As a practical basis for estimating the precipitation intensitydirectly from the measured reflectivity factor in still air, thefollowing relation is used

Z = a1 ·Rb1 . (10)

The parameters a1 and b1 are dependent on the frequency ofthe interacting EM waves and on the rain intensity R as wellas the DSD. Furthermore, regional-dependent variations duethe rain type do exist. A number of Z − R relations wereestablished by many research efforts and are provided, forinstance, in [18]. A careful selection of the coefficients has tobe performed by considering the appropriate conditions andrespective parameters.

The power law in (10) provides an analogy to the calcu-lations of the specific attenuation using (1) given in SectionIV-C.

For a further in-depth analysis of the theoretical aspects ofbackscattering due to hydrometeors, we refer to [24]. For thecase of SAR, it can be concluded that the backscattering dueto hydrometeors is the minor effect and attenuation due to theprecipitation volume is dominating, which is supported by therecent measurements of TerraSAR-X.

It has been observed that backscattering due to precipitationcan easily enhance the backscattering about 5 dB compared tounaffected regions of the image.

E. Test Case and Comparison of SAR Data with Simultane-ously Measured Ground-based Weather Radar Data

Fig. 13 shows a comparison of two different types of imagesmeasured almost at the same time, where the image on the left-hand side was acquired with TerraSAR-X [15] strip-map modein ascending orbit over New York. The image on the right-handside displays the corresponding weather radar image measuredby a ground-based weather radar (WSR-88D) located in NewYork (Nexrad code: KOKX). The data were obtained using

the freely available Java NEXRAD viewer provided by theNational Oceanic and Atmospheric Administration, U.S. Agood agreement between visible artifacts shown in the SARimage (some of them encircled in red color) and the reflectivityplot on the right-hand side was observed. Such reflectivitymaps display the echo intensity of the transmitted radar signalsand are shown in dBZ. These maps are used to detect precipi-tation and evaluate storm structures. It is the best availablemeans to compare precipitation volumes and precipitation-induced signatures in SAR images. The reason lies in the highachievable spatial resolution and the possibility to measure atalmost the same instant of time. The red regions in the weatherradar image correspond to reflectivities up to 50, in some cases55 dBZ, which corresponds to high precipitation intensitiestypically occurring during thunderstorms. The comparison ofground-based weather radar and SAR data will be certainlyuseful in the process to derive rainintensity information fromSAR-based measurements.

V. IDENTIFICATION OF MEASURES ANDRECOMMENDATION TO COUNTERACT AND/OR ENHANCE

PRECIPITATION EFFECTS ON SAR

In this section we briefly highlight some of the possibil-ities to overcome the limitations of the SAR system underinvestigation due to adverse meteorological conditions. Theideal strategy to identify rain induced signatures would bea multi-frequency SAR system, because attenuation throughrain is wavelength dependent. Since current SAR systems arenot equipped with such a multi frequency sensor set, otherways have to be taken into consideration. Wherever available,ground based weather radar would assist in the process to iden-tify precipitation induced distortions in SAR images. However,world-wide coverage of ground based weather radars is limitedto industrialized regions such as the USA or Europe. Espe-cially over the oceanic regions no ground based weather radardata are available and so it is for the polar regions of the Earth.One simple measure to avoid or mitigate precipitation effectsare multi-temporal acquisitions over the same scene, since thestatistical probability for rain is rather low, as already shownbefore. As the look angle determines the path length of thepropagation path through the precipitation media, steep lookangles would reduce the attenuation due to rain, and converselyshallow look angles cause increased values of attenuation.The question to what extent scan-on-receive techniques arecapable to improve, respectively enhance precipitation effectswas addressed by [22], and deserves further consideration. Amitigation of the backscattering can be obtained. However,the attenuation will still be present for transmit and receivepatterns as is diagrammatically shown in Fig. 14.

VI. CONCLUSIONS

A brief introduction to the SAR imaging concept was givenfollowed by a discussion of propagation distortions and theirinfluence on the image quality. Propagation effects can bevery important and need to be considered in interpretingradar images. It has been shown that attenuation due to rain

Fig. 14: A depiction of an imaging scenario with scan-on-receive technique. The different gain in receive Rx (pencilbeam) and Tx allows for an improvement of the SNR in thepresence of rain for backscattering. However, the attenuationcan not be avoided.

is the dominating effect, together with backscattering fromprecipitation at higher frequency bands such as X- and Ka-band. Since the effects of backscattering and attenuation areinterconnected to each other it is sufficient to consider at leastthe major disturbance (=attenuation through rain) in order toflag an affected SAR image. A model to quantify propagationeffects in SAR images has been presented, which allows fora quantitative assessment of the pertinent effects. Dependingon the climatic region on Earth, the availability of a Ka-band system will vary. Assuming a 5 dB acceptance of theattenuation due to rain, which corresponds to 2 mm/h at 30◦

incidence angle for the modelled rain cell used in this paper,the availability will be better than 98 % for the Europeanregions and better than 95 % for rain-forest in Brazil. Cloudswith little water liquid content, low rain rates and homogenousdistribution will cause no or only little disturbance (visibleartefacts). The disadvantage of rain cell signatures might offerthe potential advantage to derive meteorological information.

VII. ACKNOWLEDGEMENTS

I would like to sincerely thank Prof. Madhukar Chandrafrom Chemnitz University of Technology, Chemnitz, Germanyfor the invitation to present this article at the 2nd WFMN(Wave Propagation and Scattering in Communications, Mi-crowave Remote Sensing and Navigation) conference. Further-more, I would like to sincerely acknowledge the proofreadingand constructive comments by Dr. G. Krieger as well as theprovision of Fig. 2 by Dipl.-Ing. S.V. Baumgartner.

REFERENCES

[1] W. G. Carrara, R. S. Goodman, and R. M. Majewski. Spotlight SyntheticAperture Radar: Signal Processing algorithms. Artech House, 1995.

[2] I. Cumming and F. Wong. Digital Processing of SAR Data. ArtechHouse, Norwood, MA, 2005.

[3] J. C. Curlander and R. N. McDonough. Synthetic Aperture Radar:Systems and Signal Processing. J. Wiley and Sons, New York, 1991.

[4] M. Soumekh. Synthetic Aperture Radar Signal Processing with MATLABAlgorithms. Wiley, 1999

[5] J. P. Fitch. Synthetic Aperture Radar. Springer Verlag, 1988.

Fig. 13: Test case showing a comparison of TerraSAR-X and weather radar data acquired nearly simultaneously (within thesame minute) over New York, U.S. A good agreement between the rain-cell signatures in (left) the SAR image and (right) theweather radar image can be observed. The effects are most pronounced for reflectivities of up to 50 dBZ. The SAR image wasacquired in ascending orbit direction, and the range direction corresponds with the horizontal. The look direction was fromleft to right. The vertical corresponds to the along-track direction. Image dimensions are approximately 130 km in azimuth and100 km in range direction.

[6] H. Klausing and W. Holpp, editors. Radar mit realer und synthetischerApertur. Oldenburg, München, Wien, 2000.

[7] C. Oliver and S. Quegan. Understanding Synthetic Aperture RadarImages. Sci Tech, Raleigh, NC, 2004.

[8] F. M. Henderson and A. J. Lewis, editors. Manual of Remote Sensing,Principles and Applications of Imaging Radar. Wiley, 1998.

[9] C. Elachi and J. J. van Zyl. Introduction to the Physics and Techniquesof Remote Sensing. Wiley, 05 2006.

[10] H. Mott. Remote Sensing with Polarimetric Radar. Wiley, 2007.[11] J. S. Lee and E. Pottier. Polarimetric Radar Remote Sensing. CRC Press,

2009.[12] R. F. Hanssen. Radar Interferometry: Data Interpretation and Error

Analysis. Kluwer Academics, 2001.[13] H. A. Zebker, P. A. Rosen, and S. Hensley. “Atmospheric Effects

in Interferometric Synthetic Aperture Radar Surface Deformation andTopgraphic Maps.” Journal of Geophysical Research, 102:7547–7452,1997.

[14] A. Danklmayer, E. Archibald, T. Boerner, D. Hounam, and M. Chandra.“Atmospheric Effects and Product Quality in the Application of SARInterferometry.” In: Proceedings of EUSAR 2004, Ulm, Germany. ISBN3-8007-2828-1.

[15] R. Werninghaus, St. Buckreuss, “The TerraSAR-X Mission and SystemDesign” IEEE Transaction on Geoscience and Remote Sensing, acceptedfor publication in a Special issue on TerraSAR-X; available online[http://ieeexplore.ieee.org]; in hard copy to appear in 2010.

[16] R. K. Crane, Electromagnetic Wave Propagation through Rain. JohnWiley and Sons, 1996.

[17] C. Melsheimer, Signaturen von Regen in Radaraufnahmen des Meeres.Aachen: Shaker Verlag, 1998.

[18] R. L. Olsen, D. V. Rogers, and D. B. Hodge. “The a · Rb Relation in

Calculation of Rain Attenuation,” IEEE Transactions on Antennas andPropagation, vol. 26, no. 2, pp. 318–329, 1978.

[19] H. Runge, S. Cloude, M. Eineder, A. Fusco, E. Gill, I. Hajnsek,C. Heer, F. Jochim, M. Kirschner, G. Krieger, A. Moreira, T. Nieder-stadt, K. Papathanassiou, R. Romeiser, R. Scheiber, C. Sickinger, andS. Suchandt. New Techniques for Simultaneous SAR Interferometry Eur.Space Agency, Noordwijk, The Netherlands, Final Rep. ESA Contract16100/02/NL/EC, 2003.

[20] A. Danklmayer, B. Döring, M. Schwerdt, and M. Chandra. “Assess-ment of Atmospheric Effects in SAR Images,” IEEE Transaction onGeoscience and Remote Sensing, 2009, vol. 47, no. 10, pp. 3507-3518,October 2009,

[21] Specific Attenuation Model for Rain for Use in Prediction Methods,2005. ITU-Recommendation P. 838-3.

[22] S. D’Addio, M. Ludwig, “Rain Impact on Sensitivity of Ka-bandScan-on-Receive Synthetic Aperture Radars,” in IEEE InternationalGeoscience and Remote Sensing Symposium, (IGARSS) 2008, vol 3,7-11 July 2008 Page(s):III - 1174 - III - 1177, Boston, US

[23] V. N. Bringi and V. Chandrasekar. “Polarimetric Doppler WeatherRadar, Principles and Applications”, Cambridge University Press, NewYork, 2001.

[24] T. Oguchi, ”Electromagnetic Wave Propagation and Scattering in Rainand other Hydrometeors,” Proc. IEEE, vol. 71. no. 9, pp. 1029-1078,Sep. 1983